Risk Factors Evaluation for Monitoring of Well Drilling
A.S. Grigoriev, D.V. Konoshonkin, N.V. Bukhanov, O.T. Osmonalieva and B.V. Belozerov
Event name: Saint Petersburg 2018
Session: Informational and analytical support for the Mineral Resources Base Development
Publication date: 09 April 2018
Info: Extended abstract, PDF ( 609.66Kb )
Price: € 20
The purpose of this work is to create a learning algorithm that, based on historical data about previously drilled wells, will classify drilling problem. Such decision support system will help the engineer to intervene in the drilling process and prevent high expenses due to unproductive time and equipment repair due to a problem. In order to solve this task, the wells were found in which problems were encountered. Calculations have been made on various machine learning algorithms to identify an algorithm that yields a minimum error rate. As a result of the project, a model based on gradient boosting was developed to classify the problems in the drilling process.